This tests the case where something matched *both* a regex and a non-regex
rule (or two regexes), but the (correct) regex rule wasn't winning. It should
be the case, if the code is right, that if you change the order of the rules in
your grammar, you still get the same output translations.

This test tests the use of regular expressions in the grammar.  This is an
experimental feature with an inefficient implementation in the decoder, but
there are a number of things that could be done to make it more efficient if
the technique proves useful.

To enable it, you set the Joshua parameter

  regexp-grammar = OWNER

where OWNER is the owner of one or more grammars whose rules might be interpreted as regular
expressions.
